Dysfunction of the cerebellum is associated with a wide variety of diseases that have significant morbidity yet inadequate therapy. Strokes, tumors, infection, autoimmune disease, medication side effects, and chronic alcoholism can all affect the cerebellum, as can more specific diseases of the cerebellum such as olivopontine cerebellar degeneration or cerebellar ataxia. These conditions may be associated with a host of disabling medical conditions, including gait and balance problems, slurred or slow speech, blurred or double vision, tremor, and mild dementia. Indeed, cerebellar dysfunction can cause considerable degradation in the quality of life and may also be life-shortening. There is increasing evidence that regional changes in the cerebellum can be identified with specific behavioral, motor, and cognitive deficits in patients. In order to study these relationships and then to make clinical protocols available, it is necessary to develop, and test automatic procedures for analysis of the cerebellar anatomy from magnetic resonance images. The proposed research program addresses this need. Specifically, we propose to (1) optimize imaging and tissue classification methods for magnetic resonance imaging of the cerebellum;(2) develop and evaluate automatic cerebellar labeling methods;(3) investigate and model shape differences of cerebellar regions; and (4) conduct a pilot study of regional cerebellum volumes in normal subjects and cerebellar ataxia patients. The proposed research program is novel in several ways. Tissue classification and segmentation will use multiple data sets for superresolution and will incorporate topology-preserving deformable models and kernel density prior shape models. Labeling (parcellation) will incorporate topology-preserving, multiple-atlas deformable registration using multiparametric features. Since our clinical application is to diseases associated with cerebellar atrophy, our methods must work on both normal and atrophied cerebella. Our pilot study on spinocerebellar ataxia (specifically SCA6) provides a unique opportunity to study a cerebellar-specific disease and to learn about regional atrophy and its correlation with neurological deficits. It is expected that the automatic procedures developed herein will not only facilitate large-scale neuroscientific studies of the structure/function relationships in the cerebellum, but will also provide the basis for a clinical tool to assist in treatment and management of all types of cerebellar diseases.